A Method for Measuring the Constraint Complexity of Components in Automotive Embedded Software Systems

Abstract

The rapid growth of software-based functionalities has made automotive Electronic Control Units (ECUs) significantly complex. Factors affecting the software complexity of components embedded in an ECU depend not only on their interface and interaction properties, but also on the structural constraints characterized by a component’s functional semantics and timing constraints described by AUTomotive Open System ARchitecture (AUTOSAR) languages. Traditional constraint complexity measures are not adequate for the components in embedded software systems as they do not yet sufficiently provide a measure of the complexity due to timing constraints which are important for quantifying the dynamic behavior of components at run-time. This paper presents a method for measuring the constraint complexity of components in automotive embedded software systems at the specification level. It first enables system designers to define non-deterministic constraints on the event chains associated with components using the AUTOSAR-based Modeling and Analysis of Real-Time and Embedded systems (MARTE)-UML and Timing Augmented Description Language (TADL). Then, system analysts use Unified Modeling Language (UML)-compliant Object Constraint Language (OCL) and its Real-time extension (RT-OCL) to specify the structural and timing constraints on events and event chains and estimate the constraint complexity of components using a measure we have developed. A preliminary version of the method was presented in M. Garg and R. Lai, Measuring the constraint complexity of automotive embedded software systems, in Proc. Int. Conf. Data and Software Engineering, 2014, pp. 1–6. To demonstrate the usefulness of our method, we have applied it to an automotive Anti-lock Brake System (ABS).